Showcases

Reference-video-driven camera language demonstration

Use a reference video to demonstrate Seedance 2.5's reference-to-video path: treat the reference clip as a "visual note" for camera language, letting the model generate new continuous shots following the camera movement rhythm and compositional relationships of @video1, rather than performing video editing or extension. The entry point of the topic is a visual demonstration of common camera movements—push, pull, pan, follow, and orbit—presented in sections within the same 5-second shot, making it easy for developers to compare "how reference material constrains camera position and movement" during integration. The reference clip is first generated from image-sequence-style demonstration material, then passed in as reference_video_urls; base explicitly specifies duration, resolution, aspect ratio, and audio toggle to ensure the submission snapshot is reproducible. Suitable for storyboard previsualization, comparison of advertising shot plans, and camera movement testing scenarios that require character/product consistency.

Style Camera Movement · reference-to-video · Storyboard

Model Seedance 2.5

Reference inputs

reference_video_urls

Try it yourself

Prompt

Try this model

A continuous 5-second demonstration shot, presented overall as a camera movement study chart in motion. It opens with an eye-level wide shot showing a minimalist studio set: a matte gray cyclorama, a concrete pedestal with a small ceramic vase, and soft key light from camera left. The camera then performs four clearly separated movements, each about one second long, with brief stabilization between them: a slow push-in toward the vase, a pull-back to the wide shot at a matching speed, a smooth pan from left to right across the set, and a low-angle tracking movement that orbits a quarter turn around the pedestal. The movement should be mechanically clean and clearly readable, with a steady gimbal, highly consistent framing, and no handheld shake. Neutral color grading, slight film grain, deep yet soft shadows, no text overlays, no cuts, no speed ramps. The intent is instructional: each movement must read as an independent camera operation rather than a stylized flourish.

Copy the full prompt and adapt it for your own generation.